[LEDE-DEV] merging the layerscape target
Y.T. Jiang
yutang.jiang at nxp.com
Tue Sep 27 04:39:09 PDT 2016
Hi John,
After the survey found, the error are from a old usb2.0(USB_EHCI_FSL) driver which is usually used for powerpc arch, not for layerscape ls1043ardb(arm64 arch) Soc.
The default kernel config and packages not enable the old usb2.0 driver in my patch. It should be enabled by:
[x] Select all target specific packages by default
[x] Select all kernel module packages by default
[x] Select all userspace packages by
Further investigation found, if remove the kernel patch 8041, can avoid USB_EHCI_FSL be compiled. But the patch hint, USB_EHCI_FSL will appear in other arch.
8041-usb-kconfig-remove-dependency-FSL_SOC-for-ehci-fsl-d.patch
"usb: kconfig: remove dependency FSL_SOC for ehci fsl driver "
CONFIG_USB_EHCI_FSL is not dependent on FSL_SOC, it can be built on non-PPC platforms.
...
21 config USB_EHCI_FSL
22 tristate "Support for Freescale PPC on-chip EHCI USB controller"
23 - depends on FSL_SOC
24 + depends on USB_EHCI_HCD
...
If have other better ways in dealing with the dependencies in LEDE?
Thanks & Best Regards
Jiang Yutang
> -----Original Message-----
> From: John Crispin [mailto:john at phrozen.org]
> Sent: Tuesday, September 27, 2016 2:34 PM
> To: Y.T. Jiang
> Cc: LEDE Development List
> Subject: Re: [LEDE-DEV] merging the layerscape target
>
> Hi,
>
> i just tried V5 and when building the target with all packages selected i
> run into some errors
>
> make[5]: Entering directory
> `/home/blogic/source/build_dir/target-aarch64_armv8-a_musl-1.1.15/linux-
> layerscape_64b/linux-4.4.21'
> CHK include/config/kernel.release
> CHK include/generated/uapi/linux/version.h
> CHK include/generated/utsrelease.h
> CHK include/generated/bounds.h
> CHK include/generated/timeconst.h
> CHK include/generated/asm-offsets.h
> CALL scripts/checksyscalls.sh
> CC [M] drivers/usb/host/ehci-fsl.o
> In file included from drivers/usb/host/ehci-fsl.c:39:0:
> drivers/usb/host/ehci.h: In function 'ehci_readl':
> drivers/usb/host/ehci.h:741:9: error: implicit declaration of function
> 'readl' [-Werror=implicit-function-declaration]
> return readl(regs);
> ^
> drivers/usb/host/ehci.h: In function 'ehci_writel':
> drivers/usb/host/ehci.h:768:3: error: implicit declaration of function
> 'writel' [-Werror=implicit-function-declaration]
> writel(val, regs);
> ^
> drivers/usb/host/ehci-fsl.c: In function 'fsl_ehci_drv_probe':
> drivers/usb/host/ehci-fsl.c:130:3: error: implicit declaration of
> function 'clrsetbits_be32' [-Werror=implicit-function-declaration]
> clrsetbits_be32(hcd->regs + FSL_SOC_USB_CTRL,
> ^
> drivers/usb/host/ehci-fsl.c: In function 'ehci_fsl_setup_phy':
> drivers/usb/host/ehci-fsl.c:205:4: error: implicit declaration of
> function 'clrbits32' [-Werror=implicit-function-declaration]
> clrbits32(non_ehci + FSL_SOC_USB_CTRL,
> ^
> drivers/usb/host/ehci-fsl.c:244:9: error: implicit declaration of
> function 'in_be32' [-Werror=implicit-function-declaration]
> if (!(in_be32(non_ehci + FSL_SOC_USB_CTRL) & PHY_CLK_VALID)) {
> ^
> drivers/usb/host/ehci-fsl.c: In function 'ehci_fsl_usb_setup':
> drivers/usb/host/ehci-fsl.c:276:3: error: implicit declaration of
> function 'out_be32' [-Werror=implicit-function-declaration]
> out_be32(non_ehci + FSL_SOC_USB_SNOOP1, 0x0 | SNOOP_SIZE_2GB);
> ^
> drivers/usb/host/ehci-fsl.c:293:9: error: implicit declaration of
> function 'mfspr' [-Werror=implicit-function-declaration]
> svr = mfspr(SPRN_SVR);
> ^
> drivers/usb/host/ehci-fsl.c:293:15: error: 'SPRN_SVR' undeclared (first
> use in this function)
> svr = mfspr(SPRN_SVR);
> ^
> drivers/usb/host/ehci-fsl.c:293:15: note: each undeclared identifier is
> reported only once for each function it appears in
> cc1: some warnings being treated as errors
>
> to reproduce this run menuconfig and then Global build settings --->
> [x] Select all target specific packages by default
> [x] Select all kernel module packages by default
> [x] Select all userspace packages by default
>
> John
>
>
>
> On 21/09/2016 16:21, Y.T. Jiang wrote:
> > Hi Rafał and John,
> >
> > I update the patch and pull a new requests(329), please check and
> review, thanks!
> > https://github.com/lede-project/source/pull/329
> >
> > V5 patch update summary:
> > 1.Copyrights assigned to myself.
> > 2.Introduce DEVICE_TITLE and DEVICE_PACKAGES.
> > 3.Rename patches prefix with 1xxx,2xxx...
> > 4.Refresh patches by "make target/linux/refresh V=s"
> > 5.Move default packages to DEFAULT_PACKAGES.
> > 6.Optimize Build/mk_firmware.
> >
> > Thanks & Best Regards
> > Jiang Yutang
> >
> >> -----Original Message-----
> >> From: Lede-dev [mailto:lede-dev-bounces at lists.infradead.org] On
> >> Behalf Of Y.T. Jiang
> >> Sent: Tuesday, September 20, 2016 8:28 PM
> >> To: Rafa? Mi?ecki
> >> Cc: LEDE Development List; John Crispin
> >> Subject: Re: [LEDE-DEV] merging the layerscape target
> >>
> >> Hi Rafał,
> >>
> >> Thank you for the detailed comment!
> >>
> >> Update status:
> >> prefixed with ">" --done
> >> Copyright --done
> >> make target/linux/refresh V=s --done
> >> Patches prefix with 1xxx,2xxx...refer target/linux/generic/PATCHES
> >> --done
> >> using DEVICE_TITLE DEVICE_PACKAGES...refer
> >> target/linux/bcm53xx/image/Makefile --ongoing
> >>
> >> After building and features validate, I will submit a new version
> patch.
> >>
> >>
> >> Thanks & Best Regards
> >> Jiang Yutang
> >>
> >>> -----Original Message-----
> >>> From: Rafał Miłecki [mailto:zajec5 at gmail.com]
> >>> Sent: Monday, September 19, 2016 7:55 PM
> >>> To: Y.T. Jiang
> >>> Cc: John Crispin; LEDE Development List
> >>> Subject: Re: [LEDE-DEV] merging the layerscape target
> >>>
> >>> On 19 September 2016 at 12:36, Y.T. Jiang <yutang.jiang at nxp.com>
> wrote:
> >>>> Thank you for your review and suggestion.
> >>>
> >>> Sure. One more note: please take a look at your mailer configuration.
> >>> It should keep all quotes prefixed with "> " to keep discussion clear.
> >>> https://en.wikipedia.org/wiki/Usenet_quoting
> >>>
> >>>
> >>>> -----Original Message-----
> >>>> From: Rafał Miłecki [mailto:zajec5 at gmail.com]
> >>>> Sent: Monday, September 19, 2016 4:01 PM
> >>>> To: John Crispin
> >>>> Cc: LEDE Development List; Y.T. Jiang
> >>>> Subject: Re: [LEDE-DEV] merging the layerscape target
> >>>>
> >>>> On 18 September 2016 at 14:24, John Crispin <john at phrozen.org> wrote:
> >>>>> i have just spent some time reviewing the layerscape PR [1] and
> >>>>> started a full build of it. its starting to look good and i cannot
> >>>>> see any blockers. if anyone has any hold on this please let me
> >>>>> know in the next couple of days. if i dont get any vetos i will
> merge it.
> >>>>
> >>>> I can see following Copyright line over and over:
> >>>> Copyright (C) 2016 OpenWrt.org
> >>>> Yutang: did you really sign a contract with OpenWrt that included
> >>> passing your copyrights to the OpenWrt project? If not, you should
> >>> just keep Copyrights assigned to yourself.
> >>>> I really would like assigning copyrights to projects where it
> >>>> doesn't
> >>> apply.
> >>>> [I do not sign a contract with OpenWrt indeed. I refer to some
> >>>> others target while developing/backporting layerscape, I find
> >>>> almost of targets included OpenWrt.org Copyright, so I also put it
> >>>> in my code files. Now should I replace " Copyright (C) 2016
> OpenWrt.org"
> >> with "
> >>>> Copyright (C) 2016 Jiang Yutang <yutang.jiang at nxp.com>" ? or retain
> >>>> the both copyright: "Copyright (C) LEDE project, Jiang Yutang
> >>>> <yutang.jiang at nxp.com>" ?]
> >>>
> >>> You're correct, current sources are messy about this. I'm trying to
> >>> stop adding mode incorrectly copyrighted code.
> >>>
> >>> You should only have something like:
> >>> Copyright (C) 2016 Jiang Yutang <yutang.jiang at nxp.com> for the code
> >>> you have written.
> >>>
> >>>
> >>>> What about using some generic profile only and then using
> >>>> DEVICE_TITLE
> >>> DEVICE_PACKAGES to specify modules that should be included on rootfs?
> >>>> [I will try to use the two variables.]
> >>>
> >>> Thanks! This will allow building images for customized boards with a
> >>> single "make" call. It's part of recently introduced
> >>> TARGET_PER_DEVICE_ROOTFS system. You may take a look at
> >>> target/linux/bcm53xx/image/Makefile as an example. There is only 1
> >>> subtarget, but it should give you a hint anyway.
> >>>
> >>>
> >>>> Would that be possible to split patches into accepted ones
> >>>> (backports)
> >>> and LEDE-specific ones?
> >>>> [The kernel patches: dpaa/qbman/fman/etc. it is really too big and
> >>>> interference review LEDE-specially code. I will split those kernel
> >>>> patches in folder patches-4.4 as the second, and keep the rest as
> >>>> fist LEDE-specific, what do you think about it?]
> >>>
> >>> For generic patches we have a following guide:
> >>> target/linux/generic/PATCHES
> >>>
> >>> You may try to follow this, if possible. E.g. you could use 0xxx
> >>> prefix for upstream accepted patches and some other prefix 1xxx,
> >>> 2xxx, or whatever applicable for other ones.
> >>>
> >>> It isn't a strict rule for targets, but it should make your target
> >>> easier to maintain I believe.
> >>>
> >>>
> >>>> Please refresh all target patches, right now I can see they contain
> >>>> all
> >>> these things like:
> >>>> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c index
> >>>> 4cb98aa..a8a97bd 100644
> >>>> 1.7.9.5
> >>>> [I found it have conflicts in current kernel version with two
> >>>> patches(arm64/mm related, 0060 and 0061) while backporting the
> >>>> dpaa/qbman/fman driver, but I'm unacquainted with both mm and dpaa,
> >>>> our dpaa team are engaged in do upstream work and can't help me. So
> >>>> I revert the two patch to bypass this issue temporary, I would like
> >>>> to wait for more leisure time then to thorough investigate and
> >>>> solve it.]
> >>>
> >>> I think you misunderstood me. I don't have anything against your
> >>> patches, just the format. Please call make target/linux/refresh V=s
> >>> and that will convert all your patches to the expected format :)
> >>>
> >>> --
> >>> Rafał
> >> _______________________________________________
> >> Lede-dev mailing list
> >> Lede-dev at lists.infradead.org
> >> http://lists.infradead.org/mailman/listinfo/lede-dev
More information about the Lede-dev
mailing list